Super Bowl Rundown

By Michael Gertz
Tuesday, February 5, 2013

Each week we rundown every game using our advanced play-by-play statistics to examine exactly how each game was won and what it means going forward. Our expected points added (EPA) analysis assigns a net point value gained or lost to every single play so we can see exactly on which types of plays teams excelled or failed, and ultimately where games were won. The EPA while each team was on offense gives a more representative measure of offensive efficiency than real life score by separating offensive success from advantages gained or lost by defensive stops, takeaways, and scores. View our complete stats for every play type each week on the Games page.

(10-6) Ravens 34 - 31 49ers (11-4-1)
Offensive EPA: BAL (10.5) - (7.5) SF
All things considered, both teams in the Super Bowl put together very similar performances. Both managed most of their success on called pass plays, with their biggest failures coming on lost fumbles. The Ravens were very good in net passing (16.2) thanks to great normal pass plays (18.6) and avoiding interceptions. The 49ers were similarly successful on normal pass plays (17.4), but suffered a league average loss on an interception (-2.9) and just over league average losses on 3 sacks (-4.0). All together, the 49ers net passing (10.5) was still very good but put them at over a 5 point deficit to the Ravens. An interesting note was the Ravens slightly outperforming the 49ers on normal pass plays despite San Francisco passing for more yards on less attempts. The difference in actual efficiency appears to stem from Baltimore's ability to score touchdowns while passing, as red zone yardage is typically more valuable than other yardage in terms of generating points. The 49ers just about exactly made up for the passing difference on rushing plays. The Ravens were poor on run plays (-2.5), partially due to predictably running the ball to waste time later in the game. Meanwhile the 49ers had success on the ground as usual (3.3). The teams remain about evenly matched after factoring in their worst category, turnovers on fumbles lost and failed 4th down conversions. The Ravens were much worse than league average there (-5.0) due to the Ray Rice fumble lost and the failed fake field goal combined. The 49ers were about equally poor (-5.6) due to the LaMichael James fumble lost and the end of the game failed 4th down attempt. While the teams were about equal overall on standard plays, the Ravens took the edge on special teams. Though they had a poor punting mark (-2.3) due to the 2nd half shank and big return, they more than made up for it on kickoff returns (4.2) due to the Jacoby Jones return touchdown. San Francisco on the other hand was just below league average with a negative kickoff return mark (-2.6) that is more typical since the move of kickoffs up to the 35 yard line. But that isn't to say that special teams decided the game necessarily, the return touchdown was simply one of the many plays during the game where a different outcome could have likely changed the game outcome. All told, the Ravens were about 5 points better than the 49ers on passing plays, 5 points worse on run plays, and 3 points better on special teams. And as a whole both offenses dominated the opposing defenses compared to league averages.

The Impact of the Power Outage
While this storyline may have been bigger if the 49ers had managed to win, there is an almost near conscensus media opinion that the power outage benefitted the 49ers in the Super Bowl. After all, CBS flashed a new graphic displaying the discrepancy between pre- and post- power outage stats almost every 5 minutes it seemed. The first logical note to keep in mind is that it is physically impossible to argue that the outage no doubt had an impact. The mere fact that the 49ers came back from a 17 point deficit just the previous game without an outage shows that "the 49ers played better after it" is very weak evidence to support the "outage had an impact" claim. But the real way to test the hypothesis is by comparing it to similar scenarios. If such a break in play could drastically change game momentum, we would expect to see similar breaks at halftime and between quarters having impacts, but we don't. If the unexpectedness of the outage was partly impactful, then we would still expect to see similar effects after timeouts or TV timeouts for commercial breaks, but we don't. In fact not only has no one ever successfully managed to show a statistical analysis of reasons behind momentum changes, no one has even ever been able to show that momentum exists at all. Countless studies of several sports have shown that apparent momentum can be nearly perfectly modeled by random variation, just like how a coin that lands on heads 3 times in a row has no real momentum. And if playing better than their average for a few drives doesn't make a team more likely to do so on their next drive, then the argument that other things impact that non-existent entity appears even more ridiculous. There is a small chance that the outage did impact the game 1 or 2 points in either direction, due to a player not staying stretched or a coach having more time to plan, but that impact was probably negligible and shouldn't really favor either team. In fact, people say that the outage made the Ravens more tight, but a similar gut-feeling argument could be made that Ray Rice would be more likely to fumble if he was more loose and confident rather than if he was tight. The real effect at play around the time of the outage was the proven effect of a losing team being more likely to cut a deficit than have it increase (accounting for the teams being of roughly equal quality). But this effect is also not due to player motivation, it is due to conservative coaching decisions that sacrifice points in order to waste time, such as running the ball more. This effect was likely at play in the Super Bowl, but even it probably only accounted for a couple points at most, still largely insignificant compared to the fact that the 49ers simply randomly played better during that portion of the game.

The Joe Flacco Sample Size
With the recent success of the Ravens, and Joe Flacco in particular, the annual "is Joe Flacco elite?" debate is all but settled in the eyes of media analysts. There's no debating that Joe Flacco had an incredible playoff run by any measurement. In his 4 playoff games this season, the Ravens netted 14.6 EPA on called pass plays. That's even better than the Patriots league-leading 12.8 EPA over the course of the regular season. But do 4 great playoff games mean we can expect great things from Flacco in the future, even though the Ravens finished right in the middle of the league in the same total net passing EPA category during the regular season? After all, 4 games ago the majority of analysts said Flacco wasn't elite. One leading opinion is that despite what Flacco does in the regular season, he is so clutch in the playoffs that he must be elite. But the truth is, despite the Ravens having some relative success at getting to the playoffs and winning about half their games in the playoffs, Flacco hasn't been very good in the playoffs until this year. In the 3 prior years in our records, the Ravens averaged a net 2.0 pass EPA, with 2 great games and 4 somewhat poor ones. That mark is slightly worse than the league average. So if we can't really expect Flacco to be better in future postseasons than future regular seasons, what can we expect in the regular season? Let's find some similar situations. A quick look reveals a couple of similar streaks, both of which featured 4 straight interception-free games by quarterbacks that are rarely considered elite. In weeks 11-14, Cam Newton managed 12.6 net pass EPA. Better yet, in weeks 7-10 Josh Freeman remarkably netted 15.4 pass EPA, even better than Flacco's streak. Both were against slightly worse defenses than Flacco faced, but that impact likely doesn't account for more than a point or 2 difference. Both players also finished the regular season with similar totals to Flacco, and neither streak was considered career-changing like Flacco. More generally, our data and numerous other analyses have shown that just like in-game streaks, multi-game streaks are no more telling of future success than the amount they increase a team's season long averages. And that is precisely how our ratings treat them. The course of the playoffs did increase our ratings for the Ravens pass game, but not dramatically. They moved from 16th to 11th in total net passing. Since no one is arguing that Flacco has either the best or the worst receiving options in the league, 11th puts Flacco in good but not elite territory. That said, our analyses have shown that top quarterbacks are drastically underpaid in the NFL, so while Flacco is not worth as much as Brady or Brees, the Ravens probably won't be handicapping themselves much by paying a good quarterback elite quarterback money